Fireworks AI releases Inkling, a 975B multimodal MoE model

By PulseAugur Editorial · [1 sources] · 2026-07-15 22:08

Fireworks AI has released Inkling, a new 975 billion parameter Mixture-of-Experts (MoE) model. This multimodal model, which can process text, image, and audio, is available under the Apache 2.0 license. Inkling is designed as a strong foundation for fine-tuning and is accessible through Fireworks AI's platform and the Tinker platform. AI

IMPACT Provides a powerful, open-weight multimodal foundation for developers to build upon.

RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on X — Fireworks (inference infra) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Fireworks AI releases Inkling, a 975B multimodal MoE model

COVERAGE [1]

X — Fireworks (inference infra) TIER_1 English(EN) · FireworksAI_HQ · 2026-07-15 22:08

The best base model for training isn't about someone else's benchmarks.

The best base model for training isn't about someone else's benchmarks. Excited to offer @thinkymachines' first open-weights model: Inkling. 975B MoE, multimodal, with Apache 2.0. A great foundation for fine-tuning. Available day-zero. https://t.co/vXAVqfpRX9

COVERAGE [1]

The best base model for training isn't about someone else's benchmarks.

RELATED ENTITIES

RELATED TOPICS